Spaces:
Running on Zero
Running on Zero
| title: OliveGemma | |
| emoji: 🫒 | |
| colorFrom: blue | |
| colorTo: red | |
| sdk: gradio | |
| sdk_version: 6.22.0 | |
| app_file: app.py | |
| short_description: Mediterranean & European diet recognition VLM | |
| python_version: "3.12" | |
| startup_duration_timeout: 30m | |
| # OliveGemma 🫒 | |
| OliveGemma is a 3B visual-language model for fine-grained food recognition, built on the PaliGemma-2-3B backbone with LoRA fine-tuning on 17,340 images across 216 Mediterranean & European dish categories. | |
| Upload a food image and ask a question — the model will identify the dish, list likely ingredients, describe visual evidence, and more. | |
| **Model:** [`JamesZar/OliveGemma-3B`](https://huggingface.co/JamesZar/OliveGemma-3B) · **Paper:** [OliveGemma: A 3 Billion VLM for Recognising the Mediterranean & European Diet](https://huggingface.co/papers/2608.03428) · **Code:** [GitHub](https://github.com/tsiokris/OliveGemma) |